In a GPU (Graphical Processing Unit) and the like in which a plurality of arithmetic processors is integrated, a technique called UMA (Unified Memory Architecture) that shares one memory among a CPU (Central Processing Unit) and the arithmetic processors instead of using dedicated memories is used. In a UFS (Universal Flash Storage) that is a memory device standard also has Unified Memory Extension defined therein as a similar technique.
Conventionally, data (e.g., a program executed by a CPU, held cache information, data necessary for an operation) temporarily stored in an internal memory such as RAM, out of information used in a memory device, is lost when a power supply of the memory device is shut down. Therefore, the data in the RAM is saved in non-volatile memory upon the power shutdown of the memory device. The memory device transfers the data to the RAM from the non-volatile memory, every time the data is necessary.
For example, when an operation mode is changed in the memory device, data in the RAM upon the power shutdown is transferred to the non-volatile memory. In this case, data transfer takes time, so that a preparation time for changing the operation mode increases. Accordingly, there arises a problem that the operation mode cannot be changed when a response within a time required for the data transfer is needed for changing the operation mode. Poor reaction time of a whole information-processing device (application system) using a non-volatile memory device and a host device also becomes a problem.
According to these problems, many restrictions are imposed on the memory device for changing the operation mode. Consequently, it has been demanded that the operation mode is changed in a short time in the memory device.